Big Data and AI as key enabling technologies in the TV industry.


Connectivity, on-demand consumption and multiscreen scenarios have transformed the TV business from unidirectional broadcast to a much more complex, multidirectional, and highly personalized ecosystem. Current TV services both consume and generate large quantities of data

The focus of innovation on TV has shifted from content distribution to data management.

 This is a selection of Optiva Media R&I areas of interest in the scope of AI & Big Data on TV:

  • Auto-tagging of TV content based on NLP analysis of metadata 
  • Automatic content categorization to assist in content curation and recommendation.
  • Creation of TV ontologies for semantic content analysis.
  • Tool-assisted TV curation and promotion.
  • Conceptual content modelling deriving actionable semantic descriptions of content from automatic analysis of metadata and media.
  • Recommender systems for TV services 
  • Entity matching for multiple metadata source integration

Use cases

Responsible recommendation 

There is a growing concern that content recommendation algorithms in TV services, particularly those based on deep learning, suffer from a lack of explainability about the latent aspects guiding the recommendations end-users are provided with.

Current recommendations favour mainstream TV content and dismiss an important part of the existing cultural offer, typically non-mainstream content, that might otherwise receive a greater deal of attention by end-users and society in general.

Optiva Media R&I aims at developing a responsible TV recommendation system that introduces societally relevant aspects related to gender equality, inclusiveness and cultural diversity into its core modelling component. It has a double beneficial impact by letting users know about relevant aspects of the content they are offered and by unearthing potentially interesting content that might go otherwise unnoticed under the tyranny of current mainstream-oriented. 


Audiovisual content is subject to a number of regulations at the national and supranational levels aimed at promoting freedom of information, accessibility and cultural diversity while protecting audiences from inappropriate commercial and ideological activities. 

Regulations include content age rating, limitations on advertising time, product advertising and product placement, TV family hour policies and prohibition/limitation of violence, sexism, disinformation, and hate speech.

Such challenges have become increasingly harder to enforce as the size and complexity of the TV sector increase. Governmental authorities and TV providers are thus in need of technological solutions that help them supervise and enforce TV regulations.

MediaWatcher is a cloud-based service for the comprehensive AI-powered analysis of audiovisual content, to support human supervision regarding aforementioned regulations. Service may include:

  • Semi-detection of commercial breaks for airing time and length monitoring;
  • Image analysis for detection of possible product placement;
  • Detect age rating and checking it against airing time;
  • Language detection, to analyse setup/diversity;
  • Audio matching, to detect reuse of news footage and differences regarding news reporting 
  • Speech recognition, textual and sentiment analysis
  • Content tagging for subject-oriented classification.


MediaBrain provides the MediaSuite platform with intelligent metadata analysis. It automates the tasks of a TV operator in terms of programming, exploitation, and intelligent recommendation of audiovisual content. It creates new possibilities of service differentiation through the improvement of:

  • User experience.
  • Content discovery.
  • Content curation: recommendation and promotion.
  • Storytelling: Meta-generation of information associated with the main content: promotional images, trailers, advanced synopses…

The intelligent metadata analysis in a Big Data environment using AI technologies and natural language processing (NLP) all built in a cloud, assists the operator in the editorial work and automates:

  • Enrichment of the metadata with external sources.
  • Auto-tagging: Automatic extraction of keywords.
  • Meta-tagging: Thematic and personalized categorization tags.
  • Assistance in the creation of taxonomies.
  • Automatic recommendation based on the user preferences and themes thanks to NLP technologies. 2021 © All rights reserved